[Met_help] stat_analysis VGRD question (UNCLASSIFIED)

[John Halley Gotway]

Barb,

That's a pretty detailed set of questions!  I'm not sure I understand everything that's going on with UGRD/VGRD/UGRD_VGRD.  Based on your question and the UGRD/VGRD/UGRD_VGRD output lines in the
METv2.0 test scripts, I'm guessing that we're not setting the FCST_VAR and OBS_VAR columns the way we intend to.  I'll take a closer look at that.

In the meantime, here's a couple things you should know...

(1) When running Stat-Analysis with a configuration file, you can specify multiple values for "obs_var" by setting using commna-separated list.  However, when running a job on the command line, to use
multiple settings you just pass the "-obs_var" command line argument multiple times.  So in your case, "-obs_var VGRD -obs_var UGRD_VGRD" would have the desired effect ("-obs_var VGRD,UGRD_VGRD" won't
work).

This is mentioned in the comments of the Stat-Analysis configuration file...

//    Job command FILTERING options to further refine the STAT data:
//       Each optional argument may be used in the job specification multiple
//       times unless otherwise indicated. When multiple optional arguments of
//       the same type are indicated, the analysis will be performed over their
//       union:

(2) Yes, we know the output lines are very long!  But you should be able to configure your text editor to turn OFF word wrapping so that it all shows up on one long line - without having to open it
with Excel.

For example, I typically use the "kwrite" editor.  In kwrite, you turn off word wrapping in Settings->Configure Editor, and on the "Appearance" menu tab, I make sure that "Dynamic Word Wrap" is not
selected.  And in the "vim" editor, you can enter the command "set nowrap" to turn off word wrapping.  So you should be able to configure your editor to turn off word wrapping.

Also, if you'd like to use a spreadsheet on your Linux workstation, you may want to consider using OpenOffice.  It's very similar to Excel, is free, and would prevent you from having to FTP things around.

I'll look more closely at the UGRD/VGRD/UGRD_VGRD settings and let you know what I find.

John


Sauter, Barbara (Civ, ARL/CISD) wrote:
> Classification:  UNCLASSIFIED 
> Caveats: NONE
> 
> Hello,
>      I am running stat_analysis, and the only output I care about for
> now is ME, MAE, and RMSE. I am running a script without any
> StatAnalysisConfig file, based on -job aggregate_stat -line_type MPR
> -out_line_type CNT.
>      I take it the VGRD parameter is different from other parameters
> since it is used for vector calculations.  Is it appropriate to compare
> fcst_var VGRD with obs_var UGRD_VGRD?  
>      For the surface comparisons, for other parameters I just specified
> -fcst_var and stat_analysis worked fine with no -obs_var specified.  For
> VGRD, it would not work until I added -obs_var VGRD to my job line.
>      This seems to work for -obtype ADPUPA (although maybe I am not
> using all the observations that I could be using?)  For -obtype VADWND,
> I am aggregating statistics separately for two different -obs_lev
> choices of P775-625 and P875-775.  I get results for UGRD at both
> levels.  For VGRD, I get results at one level but not the other one if I
> specify -obs_var VGRD.  I get the opposite if I specify -obs_var
> UGRD_VGRD.  Although the User's Guide says the user may specify a
> comma-separated list of forecast and observation variable types, I can't
> get it to work by including -obs_var VGRD,UGRD_VGRD.
>      I have 25 .stat files, and have not looked at the MPR lines for
> each of them for -fcst_var VGRD since it is very time-consuming to do
> that.  I ftp the file to my pc, then open it in Excel.  Even the much
> smaller .txt files wrap around too much to be able to view them on my
> linux workstation.  Do you have a quick way to view output and have the
> column headings and data not wrap around in the window without importing
> it into a spreadsheet?   
> 
> Thank you,
> Barb Sauter
> 
> bsauter at arl.army.mil
> (575) 678-2840
>   
> Classification:  UNCLASSIFIED 
> Caveats: NONE
> 
> _______________________________________________
> Met_help mailing list
> Met_help at mailman.ucar.edu
> http://mailman.ucar.edu/mailman/listinfo/met_help